Every fix-or-build service comes in up to three modes, so the work fits how you'd rather run it — hands-on, hands-off, or handed straight to your own AI. Then two clean billing models, kept deliberately separate.
From a free audit to delivered, proven work.
This is the spine of the whole model. Detection and generation automate cleanly; implementing on your property is the only part that needs access — and even that is scoped and revocable.
We generate the fix or asset — schema block, rewritten page, sitemap, meta, a content draft built from real data — and you download and install it. You (or your own AI) execute. Nothing on your side to grant, nothing to revoke. The lowest-friction, lowest-cost way in.
You grant scoped, revocable access — a scoped OAuth grant, a WordPress application password, or a temporary invite — and our agents implement the work directly on your site. When it's done, you revoke. Priced on the value of done-for-you plus live-site care, not on token counts. This is where the heavy lifting lives: you skip the install entirely.
Your AI agent calls a GMMM service as a tool over the Model Context Protocol — schema generation, an audit, a content build — and gets a structured result back inside its own workflow. Billed as credits per call. It's our AI-readiness play, practiced not preached: the free audit-as-MCP is a discovery channel, and the paid MCPs expose our proprietary workflows to any agent that needs them. Our own Hong Kong context MCP is a live example.
| Mode | How it's delivered | Access needed | Who executes | Priced on |
|---|---|---|---|---|
| ① DIY | we generate it → you download & install | none | you (or your AI) | cheapest |
| ② Done-for-you | you grant scoped access → we implement → you revoke | OAuth / app-password / temp invite | us (agents) | value, not tokens |
| ③ MCP | your AI calls our service as a tool | none | your AI, via our tool | credits per call |
We keep credits and subscriptions visually and mentally separate — they're not the same thing. The audit itself is always free.
Buy credits, burn them per use. This covers most services — audit fixes, content generation, schema, entity setup, MCP calls. Pay for what you actually run, nothing standing.
An ongoing monthly arrangement for continuous work — bespoke scraping, data licensing, managed programs. For work that never really "finishes", a subscription is the honest shape.
Because a single package number would be either misleading or padded. A small brochure site and a ten-thousand-URL store need completely different work.
① DIY: we generate it, you install it — no access, cheapest. ② Done-for-you: you grant scoped, revocable access (OAuth, WordPress app-password, temp invite) and our agents implement it, then you revoke. ③ MCP: your AI agent calls our service as a tool and gets the result inside its own workflow.
Yes — it's scoped and revocable by design. Scoped OAuth can be revoked anytime; a WordPress application password can be deleted the moment we're done; a temporary invite is removed after. You grant the narrowest access the task needs, we implement, you revoke.
Credits are bought and burned per task (audit fixes, content, schema) and suit most one-off work. A retained subscription is ongoing monthly for continuous work (bespoke scraping, data licensing, managed programs). We keep the two separate — credits are not a subscription.
Because honest scope depends on your site, so a single package price would mislead or pad. Pricing is enquiry-based: the audit is free, and we quote against your actual scope afterwards.
MCP (Model Context Protocol) is an open standard that lets an AI agent call an external service as a tool. Your agent invokes a GMMM service directly in its workflow and gets a structured result back, billed per call. It's how we make our own tools AI-ready — being the tool other agents reach for.
See where you stand, then pick the services and modes that fit. No account, no sales call.
Run my free audit →